The corroboration of a new research finding by an independent group of investigators is usually welcomed as evidence supporting the validity of the initial finding. However, when the finding is consistent over a broad range of diseases and exposures, we should also ask whether this consistency represents persistent artefact. This question is raised by the paper of Shaw et al. 1 (in this issue of Epidemiology), which supports a recent observation by Botto and coworkers 2 of an interaction between vitamins and maternal fever in the etiology of congenital malformations.
Botto et al. 2 used the Atlanta Birth Defects Case-Control study to evaluate the effects of first trimester maternal febrile illness and multivitamin supplement use on the risk of seven selected birth defects (neural tube defects, cardiac outflow tract defects, ventricular septal defects, atrial septal defects, cleft lip and palate, omphalocele and limb deficiencies). These birth defects were selected because their risk was apparently reduced by the mothers’ use of multivitamin supplements (previously reported in this study). The odds ratios with maternal febrile illness and no vitamin use were in the range of 1.5–4.4 for the seven birth defects, whereas the odds ratios for febrile illness plus multivitamin use were below 1.0 for all but two of the birth defects. The authors concluded that multivitamin use might antagonize the excess risk associated with febrile illness.
Shaw and his colleagues 1 found similar (although less dramatic) results in the California-based case-control study of congenital malformations. As in Botto’s analysis, the authors selected birth defects that were reduced among infants whose mothers took vitamin supplements containing folic acid. (In Shaw’s study, these were neural tube defects, conotruncal heart defects, cleft lip with or without cleft palate, isolated cleft palate and limb anomalies.) The overall results were similar to Botto’s: vitamins reduced the risk associated with maternal fever for all defects except isolated cleft palate.
“. . .when the finding is consistent over a broad range of diseases and exposures, we should also ask whether this consistency represents persistent artefact.”
In interpreting these data, Shaw focuses on the vitamin-fever interaction. However, his study shows a similar modifying effect of vitamins on the risk associated with maternal smoking (seen across all five categories of defects). There was no risk-modifying effect of maternal vitamins with maternal alcohol consumption. However, alcohol is generally not as strongly associated with common defects as is fever or cigarettes. This is true in the California study, where there is only a weak association, or none at all, between alcohol and the various congenital malformations. Consequently, there is not much room for a modifying effect of vitamins on alcohol risk.
Both the Atlanta and the California studies are population based, with thorough classification of birth defects. One major concern discussed by the authors of both papers is the long delay between infant’s birth and maternal interview (2 to 10+ years in the Atlanta study and 4 years in the California study). The delay leaves more opportunity for reporting bias. At first glance, it might seem unlikely that recall bias would influence two exposures in opposite directions (with mothers of affected babies more likely to report a febrile illness but less likely to report using vitamin supplements). However, case mothers are likely to consider the presence of febrile illness and the absence of vitamins (as well as smoking) as “bad events,” and may thus be biased in recalling them more often. Both studies selected case mothers who were more likely to report “no vitamins” (ie, only birth defects with lower risk in the presence of maternal vitamins were included in the two studies). Those mothers may also have an increased tendency to report “febrile illness” and “smoking.”
The consistent effect of vitamins on both fever risk and smoking risk across groups of congenital malformations could be evidence for the modifying effect of vitamins on these risk factors. This result could also be interpreted as a persistent bias, exacerbated by the long recall time. Therefore it might be informative to see whether the finding also holds for studies with shorter recall time. In a recent Danish case-control study 3 of cleft lip and palate, information on maternal exposures was obtained by interview within the first weeks after birth. Data on smoking was also collected by the midwife before delivery. Some 96% of eligible cases and 94% of controls participated. The Table shows the same patterns in the Danish data as reported by Shaw and by Botto. Mothers’ use of vitamins in the first trimester seems to attenuate the association between maternal smoking and facial clefts. Although this result does not rule out recall bias, it does suggest that the long time lags between exposure and data collection in previous studies are not to blame.
Yet another explanation for the apparent protective effect of vitamins could be that vitamin consumption is a marker for a generally healthy lifestyle. It is conceivable that unknown lifestyle factors associated with vitamin consumption might reduce the risk of a broad range of birth defects. The ongoing longitudinal studies of 100,000 pregnancies in Denmark 4 may help to clarify this question and the others raised here.
For the meantime, however, it is not clear whether the apparent modifying effect of vitamins on risk factors for birth defects is a persistent reporting bias, a marker for a general healthy lifestyle, or a true consistent effect across a range of exposures and congenital malformations. We can only hope for the latter, which could provide the basis for a major step forward in the prevention of congenital malformations.
About the Author
KAARE CHRISTENSEN started his research career in reproductive epidemiology, and this remains one of his main research topics. However, as time has passed and his soccer abilities have waned, he has taken an interest in aging. He is now heavily engaged in interdisciplinary research on aging, combining methods from epidemiology, genetics and demography. He is Professor of Epidemiology at the University of Southern Denmark with a research appointment at Duke University.
1. Shaw GM, Nelson V, Carmichael SL, Lammer EJ, Finnell RH, Rosenquist TH. Influence of maternal periconceptional vitamin use on associations between selected factors and congenital anomalies. Epidemiology 2002; 13: 625–630.
2. Botto LD, Erickson JD, Mulinare J, Lynberg MC, Liu Y. Maternal fever, multivitamin use, and selected birth defects: evidence of interaction? Epidemiology 2002; 13: 485–488.
3. Christensen K, Olsen J, Nørgaard-Pedersen B, et al
. Oral clefts, transforming growth factor alpha gene variants, and maternal smoking. A population based case-control study in Denmark 1991–1994. Am J Epidemiol 1999; 149: 248–255.
4. Olsen J, Melbye M, Olsen SF, et al
. The Danish national birth cohort – its background, structure and aim. Scand J Public Health 2001; 29: 300–307.